Is AI actually worthwhile?

A number of large IT companies are investing billions in AI. This includes developing/training the programmes, building data centres and building power generation and cooling systems for the data centres.

Where is the payback for this?

Will there be sufficient paying business for more than one general AI system?

There are specific areas of machine learning that are beneficial such as medical diagnosis and quality control but these are not general purpose applications.

As I understand the only ‘profit’ that general AI can bring is a reduction in human jobs and salaries. The AI solution must be significantly cheaper than the current human solution including  background costs such as system support and upgrades.

Who takes the responsibility for problems and mistakes? If one of my staff makes a mistake that ends up as my problem. If the AI system makes a mistake that remains my problem but there is nothing I can do about it. How do you ‘un/retrain’ a LLM?

Is AI (other than certain very specific versions) a solution looking for a problem?

Is this the next .com bubble?

Parents
  • I think there are some underpinning assumptions in the setup of the discussion that aren't as real as suggested, in a "chicken/egg; which comes first" sort of way. 

    The idea that payback and profit (for whom) are at the core of our social existence is a narrative that has been adopted from the financier viewpoint, not a parents raising the future generations viewpoint.

    Yes AI (like many other 'next big thing' rushes) will be a bubble, just like tulip mania, guano (South Sea bubble), railway mania (funded by slave reparations?), Luddites, coal mining (end of), [and your .com bubble] etc.

    There will be plenty of 'rubbish Jobs' [see Graeber] generated by AI, just as there were many similar 'inefficient' jobs in all the previous economic growth cycles since 'hunter/gatherer' was an elite job.

    Defaulting to 'human by default' for road sweepers and janitor’s assistant assistant probably isn't right either. In many ways, AI is just a better search engine, spell correctors and next word generator. It won't replace humans (how to recycle those spare/waste/no-longer-useful specimens?)

    Making predictions about the future is easy, it's getting them right that the problem (Like winning any 'championship' Wink ).

Reply
  • I think there are some underpinning assumptions in the setup of the discussion that aren't as real as suggested, in a "chicken/egg; which comes first" sort of way. 

    The idea that payback and profit (for whom) are at the core of our social existence is a narrative that has been adopted from the financier viewpoint, not a parents raising the future generations viewpoint.

    Yes AI (like many other 'next big thing' rushes) will be a bubble, just like tulip mania, guano (South Sea bubble), railway mania (funded by slave reparations?), Luddites, coal mining (end of), [and your .com bubble] etc.

    There will be plenty of 'rubbish Jobs' [see Graeber] generated by AI, just as there were many similar 'inefficient' jobs in all the previous economic growth cycles since 'hunter/gatherer' was an elite job.

    Defaulting to 'human by default' for road sweepers and janitor’s assistant assistant probably isn't right either. In many ways, AI is just a better search engine, spell correctors and next word generator. It won't replace humans (how to recycle those spare/waste/no-longer-useful specimens?)

    Making predictions about the future is easy, it's getting them right that the problem (Like winning any 'championship' Wink ).

Children
No Data